Introduction To Numerical Optimization & Usage of Subroutine EVOP

1 Introduction

In all branches of science, engineering and economics it is required to estimate certain quantities so that the system, device or plan not only meets a required set of specifications but is also the best in terms of certain figure of merit generally called the goodness factor of a design. This is specially so in engineering design where it is required to estimate a set of quantities so that the end product or system meets a given set of specifications in best possible manner (ie most economic design, least weight design, least volume design, six-sigma design etc or a combination thereof). These specifications apply constraints to the choice of the design variables. In practice the specifications are such that, in general, there are infinite numbers of strategy capable of meeting the specifications. The problem then arises in ascertaining which way is the most suitable or desirable. Further, as it is not possible to generate infinite numbers of plan, it is essential to know the technique that will enable direct realization of the most desired strategy in least possible effort.

2 Course Content

2.1 Two anchor problems — most economic design of a (i) grain silo, and (ii) power supply filter

2.2 General optimization problem

2.3 Desirable features of a modern optimization algorithm

2.4 Usage of optimizer EVOP (Evolutionary Operation) in solving:

(a) Unconstrained Objective Function

(b) Constrained Objective Function

(c) Global Optimization

(d) Objective Functions Containing a Mix of Integer, Discrete and Continuous Variables

2.5 General Discussion

3 Duration

The course will have a total duration of 10 hours, split in 5 days and 2 hours per day. Initially, the course will comprise of lectures conveying the basic ideas behind computer based optimization. Emphasis will then gradually shift on usage of PCs to gain hands-on experience in solving varied optimization problems. Knowledge of computer language FORTRAN is desirable but not essential. A diskette containing the optimizer EVOP and the test problems will be provided. This optimizer is also suitable for plant optimization in real-time.

4 Who Should Attend

Audience: Virtually from any pure and applied discipline. Optimization is addressed in all spheres of human enterprise from natural sciences and engineering of whatever discipline, through economics, econometrics, statistics and operations research to management science. Practitioners of mathematical programming requiring global optimization methods in diverse technological application areas such as, nucleonic, mechanical, civil, structural, electrical, electronic, chemical engineering, high performance control systems, fuzzy-logic systems, metallurgy, space technology, chip design, networks and transportation, databases, image processing, molecular biology, environmental engineering, finance and other such quantitative topics will immensely benefit from the knowledge of using global optimizer EVOP.

5 Course Provider

Sayeed Nurul Ghani, PhD (Univ of London), DIC (Imperial College), CEng (UK), MIEE (UK), Six Sigma Green Belt Certified (USA)